Real-time finger motion recognition using skin-conformable electronics

Abstract

An integrated artificial synapse array and light-responsive motion sensor can be conformably attached to a finger and used to track finger motion in three-dimensional space. Interpreting and tracking finger motion in free space is of use in the development of control interfaces for augmented and virtual reality systems. One approach to create human-machine interfaces capable of accurate finger motion recognition is to use wearable sensors with integrated neuromorphic computing. Here we show that an integrated titanium-oxide-based artificial synapse array and organic motion sensor can be conformably attached to a finger and provide real-time motion recognition. The synaptic device and sensor exhibit well-defined synaptic and light-responsive electrical properties, respectively, as well as flexibility and mechanical robustness. The integrated synapses-sensor enables optical-electrical signal conversion and summation of post-synaptic current. Finger motions for time-resolved digit patterns (0-9) can be learned and recognized with an accuracy of up to 95% at varying strains and up to 100 strain cycles.